hip joint biomechanical approach

A biomechanical approach for dynamic hip joint

analysis

Lazhari Assassi and Nadia Magnenat-Thalmann

MIRALab, University of Geneva,Battelle, Building A, 7 Route de Drize CH-1227 Carouge, Geneva Switzerland

{assassi,thalmann}@miralab.ch,

http://www.miralab.ch

Abstract. Hip osteoarthritis (OA) is one of the most common formsof musculoskeletal disorders. Although, different factors were identifiedas potential causes of the laberal tear and cartilage degeneration, theexact pathogenesis for idiopathic OA is still not completely delineated.Given the crucial role of the mechanical behavior in the degenerativeprocess, analyzing the contact mechanics in the articular layers duringactivities could contribute to understanding the pathology. This paperpresents subject-specific and non-invasive methods which jointly encom-pass anatomy, kinematics and dynamics. This unique combination offersnew ways to individualize the diagnostic by using a physically-based sim-ulation of articular layers during motion. The simulation results showedthat strong deformations and peak stresses were observed in extreme hippostures. Medical experts correlated these simulation findings with thelocations of detected abnormalities. These observations strongly suggestthat extreme and repetitive stresses within the joint could lead to earlyhip OA.

Keywords: Hip osteoarthritis, physically based simulation, anatomicalmodeling, kinematics and dynamics, computer graphics

1 Introduction

The musculoskeletal system (MS) is composed of various and heterogeneous el-ements with complex geometries, mechanical behaviors and interactions. Thissystem provides form, support, stability, protection and locomotion to the hu-man body. Because of these these important functions, research into MS andand into related pathologies is of great interest. Indeed, musculoskeletal disor-ders (MSDs) are common causes of different pathologies and physical disability,affecting many people across the world [1]. With the aging population, the so-cial impact and economic burden (e.g., medical institutions and health insurancecompanies) of MSDs are becoming more and more important to the society [2].Therefore, a significant amount of effort has to be put into maintaining thefunctional capabilities of the aging population to allow them to have a betterquality-of-life. For young people the focus is on prevention in order to reduce the

2 A biomechanical approach for dynamic hip joint analysis

impact of MSDs and on the development of new tools which can provide usefulinformation for medical experts to improve medical procedures (e.g., diagnosis,surgery planning and rehabilitation).

Among these pathologies, osteoarthritis (OA), also known as a degenerativejoint disease, is the most common form of arthritis (articular disease). OA ischaracterized by the breakdown or the degeneration of the articular cartilagewhich becomes brittle and splits. Consequently, bones are uncovered and rubagainst each other, causing symptoms such as pain, muscle weakness and limita-tion of movement in the joint. The most common sites of OA include the hands,spine, hips and knees.

To understand the human joints mechanism and thoroughly investigate thedevelopment of OA, several studies were conducted. Different disciplines (e.g.,medicine, biology, biomechanics and applied sciences) are involved in the ex-change and combination of knowledge from different expertise domains. Despitethe growing advancements, limitations still exist and much work remains to bedone to better respond to the complexity of both the human anatomy and med-ical procedures.

1.1 Medical context

This study focuses on the hip joint, which is crucially important in the muscu-loskeletal system. The hip joint supports the weight of the body in both staticand dynamic postures. It allows for a large range of movement and for the trans-fer of high forces between the femur and the pelvis during daily activities [3]. Thehip joint is classified as a ball and socket joint, with the acetabulum acting as thesocket in which the spherical femur head articulates (see Fig.1). Both bone sur-faces are covered with an articular cartilage which prevents direct bone-to-bonecontact and allows a uniform pressure distribution inside the joint. Connected tothe acetabulum rim, the acetabular labrum is a fibrocartilaginous structure thatincreases the acetabulum depth, grips the femoral head and provides stabilityto the hip joint. The hip joint is moreover reinforced by ligaments [4]. Given itsrole in the MS, the hip joint is especially vulnerable to different pathologies andmostly OA. Although the frequency of hip OA increases with age, OA is notexclusively an aging process as it is also seen in younger patients [5]. In fact,the damage of the labrum or labral tear was associated with the developmentof hip OA. Studies have shown that a labral tear is frequently found in youngerpatients, while for older patient the labral tear is more often associated withchondral damage [10]. Therefore, they suppose that the degeneration processstarts by a labral tear and may lead to articular damages. In any case, differentfactors can be at the origin of hip joint damage.

Although, genetics, obesity, injury and infections were identified as marginalfactors, the abnormal joint morphologies including femoroacetabular impinge-ment (FAI) and dysplasia are considered as the most common reasons of thecartilage and the labrum degeneration [68]. Nevertheless, the exact mecha-nisms of degeneration are still unknown because the development process of this

A biomechanical approach for dynamic hip joint analysis 3

Fig. 1. Hip joint anatomy: bones and cartilage layers.

pathology generally takes a significant amount of time [9, 10]. Therefore, differ-ent hypotheses were suggested as potential factors. Some studies highlighted thephysical activities that produce high forces or stresses on the hip joint [11], whileothers focused on repetitive micro-trauma (e.g., hip dislocation) and extrememovements [12, 13]. Indeed, athletes may practice sports which are stressful forjoints (e.g., golf, hockey, football), as well as ballet dancers who perform someexcessive motions such as twisting, pivoting and hyper-abduction/extension.Therefore, they are considered as a population at a higher risk for developinghip labral tears or cartilage damage [14, 15]. This risk can be more importantfor athletes in the presence of other factors such as structural abnormalities [16,17].

Thus, various hypotheses related to dynamics and kinematics were proposedto explain the mechanisms of degeneration. The difficulty of establishing a linkbetween the causes and the degeneration of the labrum or cartilage is becausethey often remain undiagnosed for a period of time [18]. Nevertheless, thesehypotheses need to be investigated by analyzing the hip mechanics such as thelabrum and cartilages stresses during activities [9].

1.2 Background

Several biomechanical studies were conducted to assess the intra-articular con-tacts of the hip joint. These studies are classified in two categories: experimentaland computational methods. Experimental methods based on in vitro and invivo measurements have been performed on cadaver hips by using different tech-niques (e.g., miniature pressure transducers [19], pre-scaled sensitive films [20]and stereo-photogrammetry [21]) or by using pressure transducers implantedinto patients hip prostheses [22]. Direct measurements presented valuable re-sults but unfortunately these methods present some limitations. Indeed, it isevident that the mechanical behavior of cadaveric and living hip tissue will be


different. Moreover, the direct measurement is highly invasive and difficult toimplement in non-operated hips.

Nowadays, there is no direct and noninvasive method to directly assess thehip contact. Consequently, computational methods based on analytical and nu-merical models were proposed as non-invasive alternatives. Analytical modelsare based on mathematical functions [23, 24], while the numerical models arebased on Mass-Springs systems [25] or Finite Element methods [7, 26]. Com-pared to numerical models, analytical models are less accurate because theyneglect different aspects of biomechanics such material properties and cartilageslayers. Numerical models are widely used in numerous domains and were thusadopted for medical applications. Moreover, the evolution of computing power,the accessibility of high resolution data images and segmentation techniques re-constructing accurate 3D subject-specific models have contributed to the growinguse of computational models. These models were successfully used in differentapplications, such as the analysis of symptomatic and asymptomatic hips dur-ing daily activities [16, 17]. Nevertheless, the models used in these studies arenot fully subject-specific. In fact, studies exploit generic [16] or subject-specificanatomical models [17] but combine them with generic kinematical and phys-ical data resulting from others experimental studies [3]. Moreover, the studiedmovements are often artificial (e.g., variation of anatomical angles) or limited toroutine activities (e.g., walking, climbing stairs) which are characterized by lowamplitude [16, 24]. Finally, the results of these computational models are oftenpresented without clinical validation [17].

Therefore, there is a lack of studies combining subject-models (anatomical,kinematical and physical data) to analyze the hip joint during excessive move-ments. Nevertheless, the biomechanical modeling of a subject-specific hip jointis a difficult task and requires an adapted pipeline.

To address this issue, this paper presents a functional approach based onsubject-specific models to simulate the mechanical behavior of the hip jointunder excessive movement. The analysis of the deformation location and theassessment of the stress on the articular layers (cartilage and labrum) duringsuch movements will be helpful to determine whether such activities can be afactor of hip joint degeneration.

2 Functional approach

The proposed functional approach is based on non invasive acquisition modali-ties. Magnetic Resonance Imaging (MRI) is used for anatomical modeling, a mo-tion capture system (Mocap) for kinematical modeling and simulation models forphysical modeling. Techniques which have their base in computer graphics areused to reconstruct subject anatomical, kinematical and physical models. Thesemodels are used to set up the simulation model to achieve accurate physically-based simulation of the hip joint.


2.1 Anatomical modeling

Given the numerous differences that exist between individuals, the use of subject-specific anatomical models is of paramount importance to clinical diagnosis. Toreconstruct the subject models, the first step is to select medical modalities thatallow the best imaging of the structures to model. Compared to other modalities(e.g., Computed Tomography (CT)), MRI is a good choice for musculoskeletalimaging, because it offers the simultaneous examination of soft and bony tissue.The next step consists of devising the best imaging protocol to satisfy imagingand clinical constraints. This is achieved by an adapted MRI protocol basedon an adequate trade-off between the image quality and acquisition time [27].From the acquired medical images, a segmentation approach needs to be usedto identify the anatomical structures of interest. Despite the numerous studies,a universal segmentation approach has not yet been proposed, due to noise andartifacts inherently present in medical images. To overcome this problem, thesegmentation needs to be regulated by the introduction of constraints and prior-knowledge.

A segmentation approach based on robust deformable models is devised toaccurately segment bones and soft tissue of the hip [2830]. In this approach, eachmodels vertex is considered as a particle that is subjected to various internal andexternal forces. Internal forces constrain the shape evolution, while the externalforces attract the shape toward the anatomical boundaries. The segmentationevolution is then performed by the integration of a system of discrete differentialequations. A stable implicit integration scheme [31] based on a conjugate gradienttechnique is used. To avoid the inter-penetration of the different evolving shapes,collisions methods are implemented as well as post-processing techniques [32].Figure 2 shows various results (bone, cartilage and muscles) of the segmentationapproach.

Fig. 2. MRI dataset volume used to segment hip joint anatomical structures (bonesand cartilage) and leg muscles.


Once segmented, models need to be converted to volumetric meshes in orderto be used in physically-based simulations. Various methods are proposed to gen-erate volumetric meshes based on tetrahedral/hexahedral (Tet/Hex) elements.Nevertheless, Tet elements are more preferable than Hex elements for meshingcomplex geometry. Three main approaches are commonly used: an octree-basedmethod [33], advancing front method [34] and a method based on a Delaunaycriterion [35]. Despite the performance of classical approaches, the quality of theresulting meshes is not totally guaranteed, especially for complex geometries.In fact, a large number of Tet elements (potentially with a low quality such asslivers) is often generated.

To generate Tet meshes, meshing approaches based on 2D and 3D deformablemodels are proposed in this work. For models which can be approximated bytheir medial surface (MS), a 2D deformable medial-axis based approach [36,37] is used. This approach exploits the thickness information included into themedial surface to generate Tet meshes. For more complex shapes, where the MSis not easily computed, a 3D deformable models approach [38] based on octreesubdivision with a body centered cubic lattice [39] of the surfacic mesh is used.These approaches produce Tet meshes of satisfactory quality (with respect tothe dihedral angle and aspect ratio) and complexity (low number of Tet) whichensure the simulation accuracy and stability (see Fig.3).

Fig. 3. a) Tet meshing of the femoral cartilage based on the use of the 2D deformablemedial surface approach where color present the thickness of the model. b) Tet meshof Acetabulm cartilage and Labrum generated by the 3D deformable models approach.

To set up the mechanical model, appropriate mechanical properties andboundary conditions are assigned to mesh elements. These parameters are de-fined according to the tissue properties and their attachments.


2.2 Kinematical modeling

The description of the skeletal system movement involves the definition of specificsets of axes for each bone segment [40]. This is achieved by setting a geometricrule that constructs the axes by using selected anatomical landmarks defined onthe reconstructed 3D surface of the subjects hip and femur bones. The samebone models are used to compute the hip joint center (HJC) position [41].

The subjects movements are then recorded with an optical motion capturesystem (Vicon MX 13i, Oxford Metrics, UK) using 8 infrared cameras, samplingat 120 Hz and tracking markers in a 45.3 m3 measurement volume (3.6x4.2x3m) (see Fig.4). The set of spherical markers (7 mm) are placed according toan appropriate protocol to ensure their visibility to the cameras [20]. Unlikeother motion acquisition devices (e.g., intra-cortical pins [42], external fixators[43], fluoroscopy [44]), the optical system is not invasive and allows the record-ing of larger ranges of motion. However, due to muscle activities and inertialmovements, the skin markers move over the underlying structures. This rela-tive movement represents an artifact, typically referred to as soft tissue artifact(STA) [45]. Consequently, rigid motion of the bone segment cannot be robustlyestimated from the markers trajectories. Correcting these errors is thus necessaryfor clinical relevance.

To minimize STA, a nonlinear optimization algorithm [46] is used to find,for each segment and for each frame of movement, the best rigid transformationthat minimizes the error made globally on all the markers. Since it was observedthat joint dislocation may occur due to STA, kinematic constraints allowingsome shifts at the joint are also applied (see Fig.4 (c) and (d)). The proposedapproach [47, 48] ensures an accurate kinematical modeling for the hip joint [27].

Fig. 4. a) Movement recording by using the Mocap system and b) computed subjectposture. c) An Error position (dislocation) due to the STA and d) a corrected positionby applying the optimization and constraint approach.


2.3 Physical modeling

In addition to anatomical and kinematical models, the forces acting on the hipjoint as well as a simulation model are needed to achieve physically-based sim-ulation of the hip joint.

Hip Loads estimation: Hip forces or loads were measured in the literatureexperimentally by using telemetric implants [3, 49]. Unfortunately this in vivomethod cannot be used in a non operated hip. Moreover, the resulting datacannot be considered as subject-specific. In fact, measured forces concern agedpersons who underwent hip replacement surgery (cartilage and labrum removed).Finally, the studied movements are limited to routine activities which are notuseful in our case study.

To overcome this problem, a neuromuscular simulation [5052] is exploited asan alternative. This kind of simulation offers the possibility to estimate internalparameters (e.g., muscle activations, forces) by analyzing the subject kinematicsand kinetics during the performance of activities. Such simulations were used indifferent applications like gait analysis [53], simulation of neuromuscular abnor-malities [54], or design of ergonomic furniture [55]. In neuromuscular simulations,several sets of data measured experimentally (e.g., motion capture, force platesand Electromyography (EMG)) are exploited into a specific process [56].

In this study, a neuromuscular model is adopted [57] to analyze the dynamicsimulations of movements. A specific pipeline is required to estimate forces act-ing on the hip joint. The first step consists of scaling the generic model to matchthe anthropometry of the subject-specific anatomical model. The achieved scal-ing is based on a hybrid method using measured data resulting from differentapproaches (3D body scan model [48], 3D anatomical models, MRI data and ini-tial marker positions). These sets of data are combined and processed to realizean anisotropic scaling by calculating the scaling factors for each part of the body.From the resulting model and motion capture data, the joint coordinate values(e.g., joint angles) that reproduce the subject movement (markers positions) arecalculated by using an inverse kinematic (IK) approach (see Fig.5).

To complete the process, the ground reaction force (GRF) is required. In ourcase, a computational method based on Newtonian analysis is used to replace theunavailability of the force plates data. A dynamic 3D model using the motioncapture data and the scaling model data (body segment weights) is used to esti-mate GRF. This approach is conceivable due the nature of studied movements.Indeed, the studied movements are characterized by the static foot position ofone leg and the air position of the other leg (e.g., arabesque movement), or staticposition of both feet (e.g., bending movement). Estimated GRF FLG of the legacting on the ground is expressed as:

i

Fi = FLG +m.g =i

mi.ai (1)

where m is the subject mass, g denotes the gravity vector, mi and ai are themasses and the accelerations of the body segments, respectively.


Based on kinematical data, the body segments velocity vi and accelerationai are computed from their mass center positions pi:

vi = (pi+1 pi)/t (2)

ai = (vi+1 vi)/t (3)

where t denotes the time step.The force contact point pLG is finally computed by using the moment equa-

tion: i

Mi = FLG pLG +m.g pmc =i

mi.ai pi (4)

where pmc =

imi.pi/m is the mass body center.The output of IK and the computed GRF are used as input in an inverse

dynamic (ID) procedure to compute muscle activation, which are involved in theproduced movement. Finally, the results of the ID step are used in the analyzeprocedure step to compute forces acting upon joints. The resulting forces areexported from the neuromuscular coordinate system to the hip joint coordinatesystem which is used in the physically-based simulation.

Fig. 5. IK and ID steps of the neuromuscular simulation: Computation of the move-ment with estimated GRF and muscles activation presented by curves.

Simulation model: A physical simulation model is required to compute the de-formation of the mechanical objects. However, different criteria should be takeninto account to achieve an accurate simulation, which faithfully reflects the me-chanical behavior of biological tissues. Indeed, biomechanical constraints suchas the nonlinearity, large displacements and deformations of soft tissue have to


be considered. Different models based on mass-spring systems [25], the FiniteElement Method [16] and Finite Volume Method [58] were used to simulate thedeformations of soft tissue.

To simulate the mechanical behavior of deformable objects, our simulationmodel [37, 38] is based on a fast 1st-order Finite Element system implementa-tion, which offers a good trade-off between accuracy and computation speed.This model based on a particle-system representation, allows for the accuraterepresentation of anisotropic nonlinear viscoelastic deformation models and isparticularly well suited for modeling the behavior of highly deformable materi-als. Thanks to its lumped mass approximation, such models can be integratedwith high-efficiency numerical integration methods typically used in particle sys-tems, as well as easy and efficient integration of collision effects and geometricalconstraints. Moreover, an efficient numerical integration technique is used to pro-vide good performance in the computation of these mechanical models, both inthe context of dynamic animation and quasi-static relaxation. Concretely speak-ing, the different techniques considered to build an efficient simulation model are[5961]:

A fast 1st-order Finite Element implementation system for non linear be-havior.

3D specific improvements of the Co-rotational element transformation whichis appropriate to simulate anisotropic and isotropic materials and allow ac-curate computation of the large deformations.

Pseudo-Dynamic Stop-and-Go relaxation for fast convergence for large dis-placements.

Modeling elasticity strain-stress relationships with polynomial formulationsfor simple and efficient modeling of the non-linear material behavior.

Efficient collision processing techniques based on incremental computationmethod.

The developed simulation model is implemented in a framework offering anadequate compromise between efficiency and versatility. The accuracy of the me-chanical model has been validated through simulation comparisons with publiclyavailable Finite Element packages (FEBio [62], SOFA[63] and Code-Aster [64],and has shown to offer similar accuracy (see Fig.6). Meanwhile, computationtimes are kept very low (a few seconds per frame) thanks to ad-hoc tuning ofthe numerical integration parameters, as well as to a specific handling of colli-sions which ensures high simulation stability.

Finally, the resulting aforementioned models (meshes, kinematics, loads, etc.)are used as input for the simulation model to analyze the mechanical behaviorof the subjects hip joint.

3 Clinical Analysis

To validate the simulation results, a clinical study based on morphological andradiological analysis was performed by medical experts. To eliminate the typical


Fig. 6. Accuracy comparison between the developed model and available FEM models.Computation of deformation by applying loads on a) a simple object, b) 2 objects andc) a generic hip joint.

abnormalities of the hip joint that could lead to hip joint degeneration, a mor-phological analysis was performed to evaluate the prevalence of the subjects hipjoint. To this end, standard morphological measurements were performed [65,8] (see Fig.7). The first measurement consists of computing the depth of theacetabulum. If the acetabulum is too deep, the excessive over-coverage of thefemoral head by the acetabulum causes abutment against the acetabular rim.The depth is defined as the distance in mm between the center of the femoralhead (O and the line ARPR connecting the anterior (AR) and posterior (PR)acetabular rim. The value is considered as positive and normal if O is lateralto the line AR PR. The second measure related to the femur geometry isthe femoral alpha () neck angle. A non-spherical head damages the articularcartilages by abutting the acetabulum rim. The angle is defined by the angleformed by the line O O connecting the center of the femoral head (O) andthe center of the femoral neck (O) at its narrowest point; and the line O Pconnecting O and the point P where the distance between the bony contour ofthe femoral head and O exceeds the radius (r) of the femoral head. Deviationfrom the normal geometry is usually associated with larger angles (> 60).

Based on subject-specific data (MRI and 3D bones reconstruction), thesestandard measurement methods were numerically implemented, improving the(subjective) reading of medical images. The dancer hip was thus analyzed, ac-cording to those 2 anatomical parameters. No morphological abnormalities weredetected and it was concluded that the measured hips have an average positivedepth (left hip: 8.16 mm, right hip: 7.89 mm) and an average angle in thenormal range (36.15 < < 54.43). The results were validated by radiologicalexperts.


Fig. 7. Standard morphological measurement based on subject-specific 3D models andMRI data. Devision of the acetabulum in 8 sectors for radiological analysis.

The same radiological experts performed consensus readings of the subjectsMR images [5]. The acetabular cartilage and labral abnormalities were assessedqualitatively. For this subject, acetabular and labral lesions were diagnosed inthe posterior part of the acetabular rim (see Fig.8). To describe the exact lo-cation of the lesions, the acetabulum was divided into 8 sectors (1: anterior, 2:anterosuperior, 3: superior, 4: posterosuperior, 5: posterior, 6: inferoposterior, 7:inferior, 8: anteroinferior), as depicted in Figure 7.

Fig. 8. Radiological analysis: Diagnosis of acetabular and labral lesions in the posteriorpart of the acetabular rim.


4 Biomechanical analysis of professional ballet dancer hip

joint

This study was conducted in collaboration with doctors from the departmentof Radiology and department of Orthopedic Surgery of the University Hospitalof Geneva and female professional ballet dancer from the ballet of the GreatTheater of Geneva.

The developed approach is used to analyze the mechanical behavior of artic-ular layers of a dancers hip joint. The choice of subject is justified by the natureof practiced movements. Indeed, several dance movements such as Grand-Plie(bending), Circumduction, Arabesque, Developpe-Lateral (lateral leg bench) andDeveloppe-Avant (forward leg bench) require intensive hip flexion and/or abduc-tion with rotation. Given the subject feedback, such movements are recognizedas excessive. Therefore, the daily practice of these exercises is assumed to bea potential cause which can contribute to an early hip osteoarthritis for thesubject.

Subject-specific kinematics and kinetics of the simulated movements are dis-cretized into several frames. Femoral kinematics and joint contact forces areexpressed according to standard hip joint anatomical axes (6 degrees of free-dom) with origin located at the center of the femoral head [40]. The 3 rotationangles and 3 load components are defined along these three anatomical axes (seeFig.9).

Fig. 9. Loads (Blue axis) and movement (Flexion/Extension (Y: green axis), Adduc-tion/Abduction (X: red axis) and Internal/External rotation (Z: yellow axis))are ex-pressed in hip joint anatomical axes.

The range of motion in the subject hip joint of the simulated dancing move-

ments and the estimated GRF magnitude(GRF =GRF 2x +GRF

2y +GRF

2z )


are presented in Table 1. Movements concern the leg in elevated position (exceptGrand-Plie) and are expressed with anatomical angles (Flexion/Extension, Ad-duction/Abduction and Internal/external rotation) according to the hip jointaxes. Each movement is presented with an average and a standard deviation(SD) angle. To compare the amplitude of dancing and normal movements, walk-ing angles during stance phase are presented. The GRF concerns the second leg(foot on the ground). The forces are presented with average and SD values andexpressed as a percentage of the subjects body weight (% BW).

Table 1. Range of motion of the left hip joint and estimated GRF magnitude on theright foot. Angles are reported in deg and force in newton (N).

Movement Flex/Ext() Abd/Add() Int/Ext rot() GRF=% BW (N)

Arabe. 0/0/36.1 9.8 0/0/32.1 6.6 0/0/79.5 6.4 97.4 2.0

Circum. 52.4 26.1/0/0 0/0/37.9 24.8 0/0/22.5 9.1 95.9 7.8

Dev.Av. 71.3 17.1/0/0 0/0/22.2 7.9 0/0/34.8 8.1 95.4 10.4

Dev.Lat 72.9 35.8/0/0 0/0/54.5 15.3 0/0/11.3 28.5 96.0 4.6

Gra.Pl. 70.9 49.4/0/0 0/0/8.7 1.7 0/0/1.5 7.1 48.0 3.1

Walk.S-P 30.6 10.9/0/0 0/0/12.2 3.6 0/0/10.4 6.1 95.7 8.3

As shown in Table 1, dancing involves intensive hip flexion and abduction(except the arabesque where the hip is in extension). Globally, estimated loadsdepend on movements (angles). Figure 10 shows the evolution of loads accordingto the angles of bending movement.

Fig. 10. Curves of angles and loads for Bending movement. Angles are expressed indegree and loads in %BW.

To build the biomechanical model, the mechanical properties of soft tissueare considered. Given (i) the significant difference of the Young modulus between


the labrum, articular(20 MPa), cartilage (12 MPa) and cortical bone (17GPa) [7]and (ii) the small mechanical role of trabecular bone [17], bone deformation willbe minimal (0.01-0.1%) compared to cartilage deformation [66]. Therefore, it ismore convenient to consider bones as rigid and surfacic structures to reasonablysimplify the model and considerably reduce computation times [16].

Then, the subject-specific deformable models of the soft tissue consist of twotetrahedral meshes (see Fig.3). The first mesh (18k Tet) is composed of boththe labrum and acetabulum cartilages, where the tetrahedral elements of eachcartilage are defined with mechanical properties specific to the tissue type. Thesecond mesh (7k Tet) exclusively represents the femoral cartilage. Such modeling(2 meshes instead of 3) reduces the computation of collision detection. Externalsurfaces of tetrahedral meshes are extracted to define the boundary conditionsin the simulation model: vertices of the first mesh attached to the hip bone areconsidered as fixed, while those of the second mesh attached to the femur willtransfer loads. Since soft tissues are characterized by large deformations, whichare tackled by the used simulation model, mesh elements of the labrum, ac-etabulum and femoral cartilages are parameterized with appropriate mechanicalbehavior [67, 68].

The physically based simulation exploits this model to analyze articular lay-ers (labrum, acetabular and femoral cartilage) deformations. For each frame ofsimulated movements, the contact and the peak stress are computed. The stressrefers to the stress along the direction of the maximal compression. To analyzethe intra-articular contacts and especially the labrum deformation, peak stresseson each layer are reported. Individual analysis of stresses makes it possible toquantify the load absorbed by labrum to caracterize its role in the hip jointstructure [69].

On the other hand, the simulation calculates the peak stress locations, whichare also of paramount importance in the analysis. Indeed, such examinationprovides insight into which region of labrum or cartilage is subjected to highstress during movement in order to investigate regions susceptible to be damaged.

The simulated movements showed that usually the peak stresses were locatedin the superior and postero-superior (respectively positions 3 and 4 in Figure 7)parts of the labrum and actebuar cartilage (see Fig.11 and Fig.12). Some otherparts exhibited high stress but not in the same significant way.

5 Discussion and conclusion

The morphological measurements were analyzed by a radiological expert. Theresults of morphological analysis showed that no values indicating potential mor-phological problems were reported. FAI (Cam or pincer) morphology was not ob-served, nor was possible hip dysplasia. However radiological analysis indicatedthat lesions were observed in the superior part of the labrum and acetabular car-tilage. By putting in correspondence the location of the lesions with the stressanalysis of the simulation, high stresses were located in the superior area of theacetabular rim (see Fig.12), which corresponds to the localization of diagnosed


Fig. 11. Distribution of the stress on actebular and femoral cartilage (without thelabrum). Peak stress (red color) on position 3 and 4 of acetabular rim.

lesions (see Fig8). Then, an assumption is that excessive movements may explainthese lesions of idiopathic OA. This assumption can be supported by the natureof the dancers movements.

Fig. 12. Location of the stress peak (red color) observed during simulation in thesuperior and postero-superior parts of the labrum and actebuar cartilage.

Nevertheless additional work is required to assess some simulation compo-nents in order to fully accept the results. In fact, the accuracy of the differentstages involving anatomy, kinematics and dynamics can have some impact onthe quality of the simulation. To improve the significance of the results, analysisof more subjects is planned.


6 Acknowlegment

This work is supported by the Swiss National Research Foundation (SNSF:project 200020-132584/1). We are grateful to the University Hospital of Genevaand the ballet dancers of the great theater of Geneva for their collaboration.

References

1. Arthritis foundation, http://www.arthritis.org2. Bevan, S., McGee, R., Quadrello, T.: Key findings of the fit for work europe report

on musculoskeletal disorders and work. In: Occupational Health at Work 2009. vol.6, pp. 3030. The At Work Partnership (2009)

3. Bergmann, S.G., Deuretzbacher, G., Heller, M., Graichen, F., Rohlmann, A.,Strauss, J., Duda,J.N.: Hip contact forces and gait patterns from routine activi-ties. J. Biomech. 34, 859871 (2001)

4. Standring, S.: Grays anatomy: The anatomical basis of clinical practice. 39th Ed.,Elsevier Churchill Livingstone (2005)

5. Duthon, V., Menetrey, J., Kolo-Christophe, F., Charbonnier, C., Duc, S., Pfirrmann,C., Magnenat-Thalmann, N., Becker, C.b Hoffmeyer, P.: Professional dancers hip:Correlation of clinical and mri findings. In: Swiss Med Wkly, EMH. vol. 139, pp.2324. (2009)

6. Kelly, B.T., Weiland, D.E., Schenker, M.L., Philippon, M.J.: Arthroscopic labralrepair in the hip: surgical technique and review of the literature. Arthroscopy. 21,496504 (2005)

7. Russell, M., Shivanna, K., Grosland, N., Pedersen,D.: Cartilage contact pressureelevations in dysplastic hips: a chronic overload model. J. Orthop. Surg. Res. 1,169177 (2006)

8. Tannast, M., Goricki, D., Beck, M., Murphy, S., Siebenrock, K.: Hip damage occursat the zone of femoroacetabular impingement. J. Clin. Orthop. Relat. 466, 273280(2008)

9. Macirowski, T., Tepic, S., Mann, R.W.: Cartilage stresses in the human hip joint.:J. Biomech. Eng., 116, 1018 (1994)

10. McCarthy, J.C., Noble, P.C., Schuck, M.R., Wright, J., Lee, J., Waterman, M.S.:The Otto E Aufranc Award the role of labral lesions to development of early degen-erative hip disease. Clin. Orthop. 393, 2537 (2009)

11. Pool, A.R.: Imbalances of anabolism and catabolism of cartilage matrix compo-nents in osteoarthritis. In: V.M. Goldberg K.E. Kuettner (eds.). Osteoarthritic Dis-order, pp. 247260. Rosemont, Illinois: American Association of Orthopaedic Sur-geons (1995)

12. Narvani, A.A., Tsiridis, E., Tai, C.C., Thomas, P.: Acetabular labrum and its tears.Br. J. Sports Med. 37, 207211 (2003)

13. Narvani A.A., Tsiridis, E., Kendall. S., Chaudhuri, R., Thomas, P.: A preliminaryreport on prevalence of acetabular labrum tears in sports patients with groin pain.Knee Surg. Sports Traumatol Arthrosc. 11, 403408 (2003)

14. Binningsley, D.: Tear of the acetabular labrum in an elite athlete. Br. J. SportsMed. 37, 8488 (2003)

15. Bharam, S.: Labral tears, extra-articular injuries, and hip arthroscopy in the ath-lete. Clin. Sports Med. 25, 279292 (2006)


16. Chegini, S., Beck, M., Ferguson, S.: The effects of impingement and dysplasia onstress distributions in the hip joint during sitting and walking: A finite elementanalysis. J. Orthop. Res. 27, 195201 (2008)

17. Anderson, A.E., Ellis, B.J., Maas, S.A., Peters, C.L., Weiss,J.A.: Validation offinite element predictions of cartilage contact pressure in the human hip joint. J.Biomech. Eng. 130, 110 (2008)

18. Byrd, J.W., Jones, K.S., Smith, T.F., Waterman, M.S.: Prospective analysis of hiparthroscopy with 2-year follow-up. Arthroscopy. 16, 578587 (2000)

19. Brown, T.: In vitro contact stress distributions in the natural human hip. J.Biomech. 16, 373384 (1983)

20. Macirowski, T., Tepic, R.M.S.: Cartilage stresses in the human hip joint. J.Biomech. Eng. 116, 1018 (1994)

21. Ahmad, M.C., Cohen, Z., Levine, W., Ateshian, G., Mow, V.: Biomechanical andtopographic considerations for autologous osteochondral grafting in the knee. Amer.J. Sport Med. 29, 201206 (2001)

22. Hodge, W., Carlson, K., Fijan, R., Burgess, R., Riley, P., Harris, W., Mann R.:Contact pressures from an instrumented hip endoprostheses. J. Bone and Joint Surg.71, 13781386 (1989)

23. Mavcic, B., Pompe, B., Antolic, V., Daniel, M., Iglic, A., Kralj-Iglic, V.: Mathe-matical estimation of stress distribution in normal and dysplastic human hips. J.Orthop. Res. 20, 10251030 (2002)

24. Xishi, W., Tianying, W., Fuchuan, J., Yixiang, D.: The hip stress level analysis forhuman routine activities. Biomed. Eng. Appl., Bas. Com. 17, 4348 (2005)

25. Maciel, A., Sarni, S., Boulic, R., Thalmann, D.: Stress distribution visualizationon pre- and post-operative virtual hip joint. In: Proc. Comp. Assist. Orthop. Surg.(CAOS05). pp. 298301. (2005)

26. Ahmet, C., Vahdet, U., Recep, K.: Three-dimensional anatomic finite element mod-elling of hemi-arthroplasty of human hip joint. Trends Biomat. Artif. Orga. 21, 6372(2007)

27. Magnenat-Thalmann, N., Charbonnier, C., Schmid, J.: Multimedia application tothe simulation of human musculoskeletal system: A visual lower limb model frommultimodal captured data. In: Proc. IEEE Int. Workshop in Sig. Proc., pp. 520525.(2008)

28. Schmid, J., Magnenat-Thalmann, N.: Mri bone segmentation using deformablemodels and shape priors. In: Proc. Int. Conf. med. imag. comp. comp. ass. in-ter.(MICCAI), vol. 5241, pp. 119126. Springer-Verlag, Heidelberg(2008)

29. Gilles, B., Magnenat-Thalmann, N.: Musculoskeletal mri segmentation using multi-resolution simplex meshes with medial representations. Med. Image Anal. 14, 291302 (2010)

30. Schmid, J., Kim, J., Magnenat-Thalmann, N.: Robust statistical shape models formri bone segmentation in presence of small field of view. Med. Image Anal. 15,155168 (2011)

31. Volino, P., Magnenat-Thalmann, N.: Implicit midpoint integration and adaptivedamping for efficient cloth simulation. Comput. Anim. Virt. World. 16, 163175(2005)

32. Schmid, J., Sandholm, S., Chung, F., Thalmann, D., Delingette, H., Magnenat-Thalmann, N.: Musculoskeletal simulation model generation from mri datasets andmotion capture data. In: Recent advances in the 3D Physiological Human. pp. 320.Springer, Heidelberg (2009)

33. Shephard, M., Georges, M.: Three-dimensional mesh generation by finite octreetechnique. Int. J. Num. Meth. Eng. 32, 709749 (1991)


34. Lohner, R.: Progress in grid generation via the advancing front technique. Eng.Comput. 39, 501511 (1996)

35. Alliez, P., Cohen-Steiner, D., Yvinec, M., Desbrun, M.: Variational tetrahedralmeshing. In: SIGGRAPH05. pp. 193204. (2005)

36. Assassi, L., Guillard., G., Gilles., B., Magnenat-Thalmann, N.: Volumetric meshesbased on medial representation for medical applications. In: Proc. Comp. Assist.Orthop. Surg. (CAOS07). pp. 259262. (2007)

37. Assassi, L., Charbonnier, C., Schmid, J., Volino, P., Magnenat-Thalmann, N.: Frommri to anatomical simulation of the hip joint. Comput. Anim. Virt. World. 20, 5366(2009)

38. Magnenat-Thalmann, N., Schmid, J., Assassi, L., Volino, P.: A comprehensivemethodology to visualize articulations for the physiological human. In: Cyberworlds.IEEE Computer Society. pp. 18. (2010)

39. Molino, N., Bridson, R., Teran, J., Fedkiw, R.: A crystalline red green strategyfor meshing highly deformable object with tetrahedral. In: Proc. 12th Int. MeshRoundt. pp. 103114. (2003)

40. Wu, G., Siegler, S., Allard, P., Kirtley, C., Leardini, A., Rosenbaum, D., Whittle,M., DLima, D., Cristofolini, L., Witte, H., Schmid, O., Strokes, I.: ISB recommen-dation on definitions of joint coordinate system of various joints for the reporting ofhuman joint motion- part I: Ankle, hip and spine. J. Biomech. 35, 543548 (2002)

41. Gilles, B., Kolo-Christophe, F., Magnenat-Thalmann, N., Becker, C., Duc, S., Men-etrey, J., Hoffmeyer, P.: Mri-based assessment of hip joint translations. J. Biomech.12, 12011205 (2009)

42. Benoit, D., Ramsey, D., Lamontagne, M., Xu, L., Wretenberg, P., Renstroem, P.:Effect of skin movement artifact on knee kinematics during gait and cutting motionsmeasured in vivo. Gait and Posture. 24, 152164 (2006)

43. Cappozzo, A., Catani, F., Leardini, A., Benedetti, M., Croce, U.D.: Positionand orientation in space of bones during movement: experimental artefacts. Clin.Biomech. 11, 90100 (1996)

44. Garling, E., Kaptein, B., Mertens, B., Barendregt, W., Veeger, H., Nelissen, R.,Valstar, E.: Soft-tissue artefact assessment during step-up using fluoroscopy andskin-mounted markers. J. Biomech. 40, 1824 (2007)

45. Kepple, T., Arnold, A., Stanhope, S., Siegel, K.: Assessment of a method to esti-mate muscle attachments from surface landmarks: a 3d computer graphics approach.J. Biomech. 27, 365371 (1994)

46. Lawrence, C., Tits, A.: A computationally efficient feasible sequential quadraticprogramming algorithm. SIAM J. Optim. 11, 10921118 (2001)

47. Charbonnier, C., Lyard, E., Magnenat-Thalmann, N.: Analysis of extreme hipmotion in professional ballet dancers. In: Proc. of the 10th Int. Symp. 3D Anal.Human Mov. Amsterdam, Netherlands (2008)

48. Charbonnier, C., Assassi, L., Volino, P., Magnenat-Thalmann, N.: Motion studyof the hip joint in extreme postures. The Visual Computer. 25, 873882 (2009)

49. Park, S., Krebs, D., Mann, R.: Hip muscle co-contraction: evidence from concurrentin vivo pressure measurement and force estimation. Gait and Posture. 10, 311322(1999)

50. Delp, S., Loan, J., Hoy, M., Zajac, F., Topp, E., Rosen J.: An interactive graphics-based model of the lower extremity to study orthopaedic surgical procedures. IEEETrans. Biomed. Eng. 37, 757767 (1990)

51. Damsgaard, M., Rasmussen, J., Christensen, S., Surma, E., de Zee, M.: Analysisof musculoskeletal systems in the anybody modeling system. Simulation ModellingPractice and Theory. 14, 11001111 (2006)


52. Erdemir, A., McLean, S., Herzog, W., van den Bogert, A.: Model based estimationof muscle forces exerted during movements. Clin. Biomech. 22, 131154 (2007)

53. Piazza, S., Delp, S.: The influence of muscles on knee flexion during the swingphase of gait. J. Biomech. 29, 723733 (1996)

54. Fox, M., Reinbolt, J., Unpuu, S., Delp, S.: Mechanisms of improved knee flexionafter rectus femoris transfer surgery. J. Biomech. 42, 614619 (2009)

55. Rasmussen, J., de Zee, M.: Design optimization of airline seats. SAE Inter. J.Passenger Cars-electronic and electrical systems. 1, 580584 (2008)

56. Sandholm, A., Pronost, N., Thalmann, D.: Motionlab: A matlab toolbox for ex-tracting and processing experimental motion capture data for neuromuscular simu-lations. In: Proc. the Second 3D Physiological Human Workshop (3DPH). vol. 5903,(2009)

57. Delp, S., Anderson, F., Arnold, A., Loan, P., Habib, A., John, C., Guendelman,E., Thelen, D.: Opensim: Open-source software to create and analyze dynamic sim-ulations of movement. IEEE Trans. Biom. Eng. 54, 19401950 (2007)

58. Irving, G., Teran, J., Fedkiw, R.: Invertible finite elements for robust simulationof large deformation. In: ACM SIGGRAPH04. ACM Press, vol. 131, pp. 131140.(2004)

59. Volino P., Magnenat-Thalmann, N.: Implicit Midpoint Integration and AdaptiveDamping for Efficient Cloth Simulation. Comp. Anim. Virt. Worlds. 16, 163175(2005)

60. Volino P., Magnenat-Thalmann, N.: Stop-and-Go Cloth Draping. Vis. Comput. 23,669677 (2007)

61. Volino P., Magnenat-Thalmann, N., Faure, F.: A Simple Approach to NonlinearTensile Stiffness for Accurate Cloth Simulation. In: ACM Trans. on Graph.. ACMPress, vol. 28, pp. 105116. (2009)

62. Finite Element Software: FEBio, http://mrl.sci.utah.edu/software.php63. SOFA:Simulation Open-Framework Architecture, http://www.sofa-framework.

org

64. Finite Element Software:Code-Aster, http://www.code-aster.org65. Pfirrmann, C., Mengiardi, B., Dora, C., Kalberer, F., Zanetti, M., Hodler, J. : Cam

and pincer femoroacetabular impingement: Characteristic mr arthrographic findingsin 50 patients. J. Radiol. 240, 778785 (2006)

66. Dalstra, M., Huiskes, R., Van-Erning, L.: Development and validation of a three-dimensional finite element model of the pelvic bone. J. Biomech. Eng. 117, 272278(1995)

67. Ferguson, S., Bryant, J., Ito, K.: The material properties of the bovine acetabularlabrum. J. Orthop. Res. 19, 887896 (2001)

68. Park, S., Hung, C., Ateshian, G.: Mechanical response of bovine articular carti-lage under dynamic unconfined compression loading at physiological stress levels.Osteoart. Cart. 12, 6573 (2004)

69. Henak, C.R., Ellis, B.J., Harris, M.D., Anderson A.E., Peters C.L., Weiss J.A.:Role of the acetabular labrum in load support across the hip joint. J. Biomech. 44,22012206 (2011)

hip joint biomechanical approach

Documents

hip osteoarthritis oa

common sites of oa

thedevelopment of oa

earlyhip oa

idiopathic oa

medical experts

based simulation

medical institutions